Re: Support for non-unique process labels?

Fred Hebert <[email protected]> writes:
> So there's a long-term plan to fold Rebar3 stuff into OTP, but the true
> easy underlying problem there is that we'd have to re-license a lot of
> code, or otherwise rewrite it to align the licenses with what would fit
> into OTP. Furthermore, the two maintainers (myself included) have limited
> time for it and haven't had it as part of work duties for the last 3-4
> years or so.
>
> In short, the sort of things that would improve that tooling would come
> down to "large corporations decide to adopt public projects or replace them
> and pay for commercial-level support of Erlang/OTP." Until then, a lot of
> the things you see as necessary require either a much larger community, or
> for some larger group of people to foot the bill so that the
> externalization that has taken place is covered for you.
>
> Until then these sorts of demands unfortunately fall in "whatever free time
> the 1-3 people maintaining a subset of the code have to give to random
> people", and that time has drastically shrunk across the board (not just
> rebar3/relx) with the pandemic and whatnot, and priority is given to
> breaking bugs for existing users rather than major refactors for the sake
> of future convenience.
>
> The level of support a language like Go (or Rust) would see hardly happens
> with something like Erlang, where Ericsson does provide *a lot* of
> significant improvements, but the community is historically closer to a
> bunch of commercial actors working on their stuff, with 2-3 more
> public-facing companies doing some work and releasing it from time to time.
>
> Until a bunch of people feel they'll get a well-paying job through the
> visibility of improving the ecosystem, I get the feeling it will be a lot
> harder to just get tons of volunteer work on tooling like that, where year
> after year, the comments (as seen in this thread) are that whatever you do
> just sucks too much and why don't you have the power of google thrown at it
> instead. You get used to it, but it's not attractive work.

It seems the discussion got personal, and I am sorry for that. I am
grateful for the work produced by volunteers, and I send bug reports and
patches when possible (as I did for rebar3 recently and will continue to
if you are interested).

But it is frustrating that every tentative to find solutions for
shortcomings in Erlang/OTP is met with a mix of denial, anger, and
whataboutism. Leading directly to everyone writing countless hacks and
workarounds in their own backyward because there is no point in trying
to improve things.

And at the risk of getting banned for saying that, I am fed up with the
old argument of Erlang not having the founding of Go, Rust, or any other
language. Ericsson is a 40+ billion dollars company with almost 100'000
employees. I get it, they do not owe anyone anything, and I do not mind
sending patches when I can. But it is not a tiny startup struggling to
pay a handful of employees to work on a platform they use, and they do
profit directly from the open source community, so it is a two way
street. And while I cannot know for sure about any large commercial
support contracts (this kind of information is surprisingly hard to
find), I would be really surprised if anyone was operating Erlang in
critical services (starting with telecoms) without priority support
contracts with Ericsson.

I will refrain from further comments, it is clear the thread is not
going anywhere.

-- 
Nicolas Martyanoff
http://snowsyn.net
[email protected]